TDA4VM: TDA4VM: OPENVX hangs on a pipelined graph when a node input is more than 32 input/output objects

Yongsig Do

Part Number: TDA4VM

Hi,

I patched TI's guide in psdk 7.1.

--> Link :TDA4VM: OpenVX node processing failed when using more than 32 input/output objects - Processors forum - Processors - TI E2E support forums

It looks like It's OK on the non-pipelined graph after patching.

But, the system hangs on the pipelined graph after patching when a node input is more than 32 objects.

Could you give me more patches for the pipeline working?

Best regards

Yongsig

over 4 years ago

0 Lucas Weaver over 4 years ago

TI__Genius 13795 points

Hello,

Could you please provide a log for the issue you are seeing?

Also, could you please provide information about the node that requires more than 32 objects? Which data types are being used and how many total objects are being used?

Regards,

Lucas

0 Yongsig Do over 4 years ago in reply to Lucas Weaver

Expert 1650 points

Hi Lucas,

Unfortunately, there are no error logs because the system just hangs.

The only thing that I know processing count is 0 when I entered 'p' on the terminal.

This node is for the post-processing after a couple of TIDL nodes.

Actually, we need 54 inputs for this node. The data types are as below.

- vx_user_data_object : 11

- vx_tensor : 43

Best regards

Yongsig.

+1 Lucas Weaver over 4 years ago in reply to Yongsig Do

TI__Genius 13795 points

Hello Yongsig,

Would it be possible for you to try this on the 7.3 SDK with the patch applied for > 32 input/outputs? I was told be a colleague that there have been changes in the 7.3 SDK memory map which would allow this to succeed.

Regards,

Lucas